According to https://stacks.math.columbia.edu/tag/08J5 we have for every complex $K$ and integer $a$ a distinguished triangle $$\tau_{\leq a}K\rightarrow\tau_{\leq a+1}K\rightarrow H^{a+1}(K)[-a-1]\rightarrow\tau_{\leq a}K[1].$$ If we consider the associated long exact sequence we get $$0=H^{a+1}(\tau_{\leq a}K)\rightarrow H^{a+1}(\tau_{\leq a+1}K)=H^{a+1}(K)\rightarrow H^{a+1}(H^{a+1}(K)[-a-1])=0$$ and so we should have $H^{a+1}(K)=0$, but there is no reason for this to be true. What is going wrong here?


I'm a newb with derived categories, so this may be wrong:

What's the convention used for the shift functors in the stacks project? I think the confusion comes from the fact that we are using cochains in $K(\mathcal A)$.

In section 13.9, we are referred back to the following:


So, if $A^\bullet$ is a cochain complex in $\mathcal A$, the shifted complex is

$A[k]^n = A^{n+k}$


$A^\bullet: \cdots \rightarrow A^{-2} \rightarrow A^{-1} \rightarrow A^0 \rightarrow A^1 \rightarrow A^2 \rightarrow \cdots $

gets shifted to

$A[k]^\bullet: \cdots \rightarrow A^{-2 + k} \rightarrow A^{-1 + k} \rightarrow A^k \rightarrow A^{1+k} \rightarrow A^{2+k} \rightarrow \cdots$

That is, $[1]$ is a left shift !

That makes sense because the $C \rightarrow A[1]$ part of a distinguished triangle should record the map $C^n \rightarrow A^{n+1}$ in degree $n$.

Now, I believe the convention is that $H^{a+1}(K^\bullet)$ denotes the complex

$ \cdots \rightarrow 0 \rightarrow 0 \rightarrow H^{a+1}(K^\bullet) \rightarrow 0 \rightarrow 0 \cdots$,

with $(H^{a+1}(K^\bullet))^0 = H^{a+1}(K^\bullet)$ and $H^{a+1}(K^\bullet))^n = 0$ otherwise.

Then, the right shift $H^{a+1}(K^\bullet)[-a-1]$ is the complex with

$(H^{a+1}(K^\bullet)[-a-1])^n = \begin{cases} H^{a+1}(K^\bullet), & n=a+1 \\ 0, & \mbox{otherwise} \end{cases}$

so then

$H^n(H^{a+1}(K^\bullet)[-a-1])) = \begin{cases} H^{a+1}(K^\bullet), & n=a+1 \\ 0, & \mbox{otherwise} \end{cases}$

and the long exact sequence associated to that distinguished triangle is

$\scriptsize \require{AMScd} \begin{CD} \cdots H^a(\tau_{\leq a}K^\bullet) @>>> H^a(\tau_{\leq a+1} K^\bullet) @>>> H^a(H^{a+1}(K^\bullet)) @>>> H^{a+1}(\tau_{\leq a}K^\bullet) @>>> H^{a+1}(\tau_{\leq a+1} K^\bullet) @>>> H^{a+1}(H^{a+1}(K^\bullet)) @>>> H^{a+2}(\tau_{\leq a} K^\bullet) \cdots \\ @VVV @VV=V @VV=V @VV=V @VV=V @VV=V @VV=V \\ \cdots H^a(K^\bullet) @>>> H^a(K^\bullet) @>>> 0 @>>> 0 @>>> H^{a+1}(K^\bullet) @>>> H^{a+1}(K^\bullet) @>>> 0 \cdots \end{CD}$

(Sorry if that goes off your screen, I'm not sure how to make arrows shorter in amscd)

  • $\begingroup$ Indeed, you are right. Thanks! For some reason I thought it was a left shift.. $\endgroup$ – Bernie May 8 at 12:06

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.